regulation of axonal morphogenesis by the mitochondrial ...€¦ · function of efhd1 in sensory...

Research Article

Regulation of axonal morphogenesis by themitochondrial protein Efhd1Valeria Ulisse1, Swagata Dey1, Deborah E Rothbard1 , Einav Zeevi1, Irena Gokhman1, Tali Dadosh2, Adi Minis1,Avraham Yaron1

During development, neurons adjust their energy balance tomeetthe high demands of robust axonal growth and branching. Themechanisms that regulate this tuning are largely unknown. Here,we show that sensory neurons lacking liver kinase B1 (Lkb1), amaster regulator of energy homeostasis, exhibit impaired axonalgrowth and branching. Biochemical analysis of these neuronsrevealed reduction in axonal ATP levels, whereas transcriptomeanalysis uncovered down-regulation of Efhd1 (EF-hand domainfamily member D1), a mitochondrial Ca2+-binding protein. Geneticablation of Efhd1 in mice resulted in reduced axonal morpho-genesis as well as enhanced neuronal death. Strikingly, thisablation causes mitochondrial dysfunction and a decrease inaxonal ATP levels. Moreover, Efhd1 KO sensory neurons displayshortenedmitochondria at the axonal growth cones, activation ofthe AMP-activated protein kinase (AMPK)–Ulk (Unc-51–likeautophagy-activating kinase 1) pathway and an increase inautophagic flux. Overall, this work uncovers a new mitochondrialregulator that is required for axonal morphogenesis.

DOI 10.26508/lsa.202000753 | Received 23 April 2020 | Revised 4 May2020 | Accepted 5 May 2020 | Published online 15 May 2020

Introduction

During development, neurons grow axons that elongate over sub-stantial distances and branch for proper tissue innervation(Vaarmann et al, 2016). In humans, the axons of the peripheralnervous system (PNS) can reach lengths of up to 1 m (Misgeld &Schwarz, 2017). As axonalmorphogenesis is energetically demanding,it must be supported by a tightly regulated energy balance.

Axonal ATP is produced primarily in the mitochondria, which arepredominately localized inmetabolically active zones of the neuronsuch as the growth cones at the leading edge of the axon(Vaarmann et al, 2016; Sheng, 2017). Mitochondrial function is criticalto axonal morphogenesis; numerous reports have demonstratedthat mitochondrial biogenesis, localization, trafficking, and local

ATP production are all limiting factors for axonal growth and mor-phogenesis (Courchet et al, 2013; Spillane et al, 2013; Vaarmann et al,2016; Misgeld & Schwarz, 2017). However, the regulatory mechanismsthat couple axonal morphogenesis and energy supply remain poorlyunderstood. The tumor-suppressor protein liver kinase B1 (Lkb1, alsocalled Stk11) is a well-known regulator of cellular polarization inepithelia (Hardie, 2007; Shackelford & Shaw, 2009) and other non-neural tissues in Drosophila and vertebrates (Nakano & Takashima,2012). In addition, studies in nonneuronal cells have established acritical function of the Lkb1 pathway in energy homeostasismediatedthrough enhancement of mitochondrial activity, mitochondrialbiogenesis, and autophagy, as well as via a mammalian target ofrapamycin-dependent decrease in energy expenditure and proteinsynthesis (Alexander & Walker, 2011; Hardie, 2011). Studies of theneuronal function of Lkb1 in the central nervous system (CNS) initiallyrevealed its key role in establishing axon polarization and extensionthrough the activation of the synapses of amphids defective kinases(Barnes et al, 2007; Shelly et al, 2007). More recently, deletion of Lkb1in the CNS revealed that it also contributes to axonalmorphogenesis,in part through its effect on mitochondrial movement, biogenesis,and localization (Courchet et al, 2013; Spillane et al, 2013).

This study reports the discovery of a new pathway that couplesenergy homeostasis to axonal growth. In our investigation, weablated the Lkb1 gene in mice at the onset of PNS development.Lkb1-KO animals exhibited abnormal axonal growth and branchingand reduced axonal ATP production. Intriguingly, transcriptomeanalysis of Lkb1 KO sensory neurons uncovered significant down-regulation of the RNA transcript of the mitochondrial protein EF-hand domain family member D1 (Efhd1, also known as mitocalcin).Efhd1 is a calcium-binding protein that is localized to the innermitochondrial membrane (Tominaga et al, 2006). To explore thefunction of Efhd1 in sensory neurons, we generated an Efhd1 KOmouse line. Herein, we characterize these animals and demon-strate that Efhd1 regulates mitochondrial function and axonalmorphogenesis during PNS development, providing a novel link ofmitochondrial activity and energy homeostasis to axonalmorphogenesis.

1Department of Biomolecular Sciences, The Weizmann Institute of Science, Rehovot, Israel 2Department of Chemical Research Support, Faculty of Chemistry, TheWeizmann Institute of Science, Rehovot, Israel

Correspondence: [email protected]; [email protected] Minis’s present address is Strang Laboratory of Apoptosis and Cancer Biology, The Rockefeller University, New York, NY, USA

© 2020 Ulisse et al. https://doi.org/10.26508/lsa.202000753 vol 3 | no 7 | e202000753 1 of 12

on 1 April, 2021life-science-alliance.org Downloaded from http://doi.org/10.26508/lsa.202000753Published Online: 15 May, 2020 | Supp Info:

http://crossmark.crossref.org/dialog/?doi=10.26508/lsa.202000753&domain=pdfhttps://orcid.org/0000-0003-3629-1367https://orcid.org/0000-0003-3629-1367https://orcid.org/0000-0001-9340-7245https://orcid.org/0000-0001-9340-7245https://doi.org/10.26508/lsa.202000753mailto:[email protected]:[email protected]://doi.org/10.26508/lsa.202000753http://www.life-science-alliance.org/http://doi.org/10.26508/lsa.202000753

Results

Lkb1 KO sensory neurons display normal polarization but reducedaxonal growth in vitro

To test the function of Lkb1 in the development of the PNS, we ablatedthe floxed Lkb1 gene in the mouse at embryonic day 9 (E9) using theWnt1–cre line, generating the strain henceforth referred to as Lkb1 KO(Swisa et al, 2015) (Fig S1A). We first tested the polarization of dorsalroot ganglion (DRG) neurons in vitro. After transfecting WT and Lkb1KOneuronswithmCherry- andGFP-expressing plasmids, respectively,we cocultured the differentially labeled cells. This approach elimi-nates any effects thatmay arise from technical variations between thecultures or non-cell autonomous effects (such as secreted factors).Dissociated DRG neurons at E12.5 typically exhibit polarized mor-phology with a pair of axons growing from two opposite sides of thesoma (Tymanskyj et al, 2018). Analysis of the Lkb1 KO and WT neuronsestablished that after 48 h, both cell types exhibit normal polarizedmorphology, with two axonal branches sprouting from opposite sidesof the cell body (Fig S1B and C). These results support the conclusionof a previous study that suggested Lkb1 is dispensable for axonformation/polarization outside of the cortex (Lilley et al, 2013).

We next examined the importance of Lkb1 for axonal growth in vitro.DRG explants from E13.5 embryos were grown for 5 d embedded in 3Dcollagen matrix (Fig 1A and B). α-βIII-tubulin staining of the explantsrevealed that Lkb1 KO axons were significantly shorter (50%) than thoseof the WT littermate controls (Fig 1E). Thus, unlike hippocampal andcortical neurons (Barnes et al, 2007; Shelly et al, 2007; Courchet et al, 2013),DRG sensory neurons are capable of establishing polarity in the absenceof Lkb1, but their axon growth potential is significantly compromised.

Lkb1 KO sensory neurons exhibit reduced axonal morphogenesisin vivo

Next, we assessed the role of Lkb1 in axonal morphogenesis in vivo.Limbs from E13.5 WT and Lkb1 KO embryos were stained with α-βIII-tubulin and visualized by microscopy (Fig 1C and D). Compared withWT limbs, Lkb1 KO limbs showed reduced axonal morphogenesis.Axonal morphology was quantified using NeuroMath (Rishal et al,2013). Relative to those in WT mice, limbs in Lkb1 KO mice displayeda strong reduction in the overall axonal coverage of the limb (36%)(Fig 1F) and an evenmore pronounced decrease in the total numberof axonal branches (48%) (Fig 1G). To rule out the possibility thatthese observed phenotypes are secondary to neuronal death, wecounted the number of Islet1 (a pan-sensory neuronal marker)-positive cells in DRGs of E15.5 embryos. We could not detect anysignificant difference in neuronal numbers between Lkb1 KO andWT control littermates (Fig S1D), suggesting that the axonal ab-normalities observed in the Lkb1 KO mice were not caused by celldeath. These results highlight the crucial role of Lkb1 as a regulatorof developmental axonal morphogenesis in the PNS.

Lkb1 KO neurons display reduced axonal ATP levels

As Lkb1 ablation had no effect on sensory axon polarization, wehypothesized that the reduced axonal growth observed in the KOs

in vitro and the reduced axonal morphogenesis phenotypes detected invivo may be linked to the distinct function of LKB1 in metabolic ho-meostasis (Alexander & Walker, 2011; Germain et al, 2013; Hawley et al,2003; Pooya et al, 2014). Because the Lkb1 KO exhibits distinct axonalphenotypes, we determined the ATP concentration in different neuronalcompartments using afilter culture system that allows us to differentiallyexamine the cell bodies and the axons (Fig 1H). Whereas there was nodifference observed between the ATP levels in Lkb1 KO andWT soma, theATP levelwas significantly reduced (49%) in theaxonsof Lkb1KOneuronscomparedwithWT (Fig 1I). This dramatic decrease in ATP levels in Lkb1 KOaxons was not accompanied by activation of the metabolic sensor AMPKor its direct substrate acetyl CoA carboxylase (ACC), as judged by theextent of their phosphorylation (Fig S1E–G). This suggests that Lkb1 is anonredundant activator of AMPK in sensory neurons. Taken together,these in vitroand in vivo results demonstrate that Lkb1 serves tomaintainnormal axonal development and ATP levels in mouse PNS neurons.

Lkb1-deficient sensory axons display normal mitochondriamotility

As mitochondria are the main source of axonal ATP and Lkb1controls the axonal movement of the mitochondria in corticalneurons (Courchet et al, 2013), we tested the motility of mito-chondria in sensory axons of DRG neurons from Lkb1 KO and WTmice. Lkb1 KO and WT neurons were plated on poly-D-lysine (PDL)/laminin–coated microfluidics chambers for 72 h (Maor-nof et al,2016). The mitochondria were labeled by TMRE, which localizes tomitochondria, and the cells were imaged. No significant differencein mitochondrial motility was detected (Fig S1H–J).

Lkb1 KO sensory neurons exhibit reduced expression of themitochondrial protein Efhd1

In addition to directly phosphorylating its downstream targets, Lkb1signals activation of gene transcription (Shackelford & Shaw, 2009).Therefore, to identify differential gene expression in DRGs directlyisolated from E13.5 WT and Lkb1 KO embryos, we performed tran-scriptome profiling using microarray analysis. Surprisingly, there werevery few alterations in gene expression in Lkb1 KO DRGs compared withWT controls. Most profound was the reduction in the expression ofEfhd1, a mitochondrial Ca2+-binding protein (approximately threefold)(Fig S2A and Table S1). In support, we detected a significant decrease inaxonal Efhd1 protein expression in Lkb1 KO neurons compared with WT(Fig 1J and L). To further test the connection between Efhd1 expressionand the Lkb1 pathway, we treated DRG neurons in vitro with the AMPKinhibitor compound C. After 8 h, we detected a profound reduction inaxonal Efhd1 (Fig 1K and M). We have not detected a reduction in thelevels of the mitochondrial protein TOM20 in the Lkb1 KO or compoundC–treated WT neurons (Fig S2B–E), suggesting that the reduction inEfhd1 is not due to global reduction in mitochondrial proteins.

Overall, these data show that Lkb1 pathway inhibition causesEfhd1 down-regulation in DRG neurons.

Efhd1 is required for axonal growth in vitro

To investigate the physiological function of Efhd1, we generated anEfhd1 KO mouse using the CRISPR–Cas9 technology (Fig S3A).

Efhd1 regulation of axonal morphogenesis Ulisse et al. https://doi.org/10.26508/lsa.202000753 vol 3 | no 7 | e202000753 2 of 12

https://doi.org/10.26508/lsa.202000753

Complete elimination of Efhd1 expression was confirmed byWestern blot (Fig S3B). To determine if Efhd1 is required for axonalgrowth in DRG neurons in vitro, we first cultured DRG explants fromWT and Efhd1 KO E13.5 embryos for 48 h on PDL/laminin, which ishighly permissive. Under these culture conditions, we have notobserved any effect on axonal growth (Fig S4A–C). Next, we culturedE13.5 DRGs for 5 d in 3D collagen matrix. Axons were visualized byα-βIII-tubulin staining (Fig 2A and B). Under these conditions, axonsof the Efhd1 KO neurons measured consistently shorter (18%)compared with those from WT littermates (Fig 2E). These results arereminiscent of the Lkb1 KO phenotype and are consistent withprevious studies on Efhd1 (Tominaga et al, 2006), supporting thenotion that Efhd1 is required for axonal growth in vitro.

Efhd1 KO sensory neurons display aberrant axonal developmentand increased cell death in vivo

Next, we assessed whether Efhd1 is also required for axonalmorphology in vivo. To this aim, we visualized Efhd1 KO and

littermate WT limbs of E13.5 and E14.5 embryos and quantifiedaxonal morphology by NeuroMath. At E13.5, the axonal patterns inWT and KO appeared identical (Fig S4D, E, and H). However, at E14.5(Fig 2C and D), we detected a significant 18% reduction in the overallaxonal coverage (Fig 2F) and a 19% decrease in the total numberof branches (Fig 2G). Similar results were obtained usingα-neurofilament staining (Fig S4F, G, and I).

We then investigated whether the lack of Efhd1 provokedneuronal loss. To assess the extent of cell death in our system,we stained for the pan-DRG neuronal marker, Islet1, and pro-cessed (i.e., active) caspase-3, which is an indicator for apo-ptotic cell death. No differences were noted at E15.5 betweenthe numbers of neurons and the apoptotic rates in Efhd1WT andKO DRGs (Fig S4J and K). In contrast, at E17.5, a significant de-crease in neuronal numbers (28%) was detected in Efhd1 KODRGs (Fig 2H–J), along with an increased rate of apoptotic cells(36%) (Fig 2L, M, and K).

Having demonstrated that the aberrant phenotypes in Efhd1 KOmice begin to arise during early development, we examined

Figure 1. Liver kinase B1 (Lkb1) KO embryos displayreduced axonal morphogenesis along withmetabolic abnormalities and a decrease in themitochondrial protein Efhd1.(A, B) Dorsal root ganglion (DRG) explants plated in3D collagen. Scale bar: 1,000 μm. (C, D) Limbs from WTand Lkb1 KO E13.5 mouse embryos stained with α-βIII-tubulin. Scale bar: 500 μm. (E) Quantification of theaxonal length of WT and Lkb1 KO DRGs. Axonal lengthsof eight WT and eight Lkb1 KO DRG explants werequantified by four measurements in threeindependent experiments N = 3. Graphs show means ±SEM (unpaired t test **P = 0.0071). (F) Overall axonalcoverage (total axonal length that cover the limb’ssurface). Graph shows means ± SEM (unpaired t test***P = 0.0001). (G) Total number of branching wasquantified using NeuroMath. Graph shows means ±SEM (unpaired t test ***P = 0.0001). (F, G) Eight WT (N = 8)and six Lkb1 (N = 6) KO embryos were analyzed in (F, G);data for each embryo represent the averagemeasurements of both limbs. (H) Schematic illustrationof the filter cell culture system that facilitates thepurification and biochemical analysis of cell bodiesand axons. (I) Analysis of ATP levels in WT and Lkb1 KOsoma and axons. Graph shows means ± SEM of threeindependent experiments (N = 3) (unpaired t test:soma NS, unpaired t test: axons **P = 0.0019). Values arepresented as the % of WT. (J) Immunoblot analysis ofEfhd1 expressions levels in Lkb1 KO and WT soma andaxons. (K) Immunoblot analysis of Efhd1 expression inWT DRGs treated with 20 μM compound C for 8 h. (L)Quantification of Efhd1 protein level in Lkb1 KO DRGscompared to WT. Graph shows means ± SEM of threeindependent experiments (N = 3) (unpaired t test: somaNS, unpaired t test axons **P = 0.00232). (M)Quantification of Efhd1 expression in treated WT DRGs.Graphs show mean ± SEM of three independentexperiments (N = 3) (unpaired t test: soma NS,unpaired t test axons ***P = 0.0004).Source data are available for this figure.


https://doi.org/10.26508/lsa.202000753

whether innervation in adult animals is also affected. After stainingthe hind limb glabrous skin with α-βIII-tubulin antibody (Fig 2N andO), we detected a significant reduction (33%) in the number ofterminal fibers in Efhd1 KO hind limb skin compared with the WTcontrol (Fig 2P). Overall, these data suggest that Efhd1 is required insensory neurons for axonal morphogenesis during developmentand for proper target innervation.

Efhd1 KO axons manifest lower ATP levels and shortenedmitochondria

As Efhd1 is a mitochondrial protein, we examined whether its ab-lation affected energy homeostasis of the neuron. We measured ATPlevels inWTand Efhd1 KOneurons fromE13.5 embryos and found thatwhereas no significant differences in ATP levels were observed

between the corresponding WT and KODRGs somas, the ATP levels ofEfdh1 KO axons are much lower (49%) than that of WT axons (Fig 3A).

Having established that axonal ATP is impacted by Efhd1 defi-ciency, we reasoned that mitochondria numbers or morphologymight also be affected. We examinedmitochondria number in Efhd1KOs using two approaches. First, we directly counted the number ofmitochondria at the axonal growth cones of DRG explant (seeschematic illustration in Fig 3B) after visualizing them using super-resolution microscopy. No significant difference in mitochondrialnumber was observed between the WT and Efnd1 KO DRGs (Fig 3C).Next, we quantified the total mitochondria mass in DRG neurons byconducting real-time PCR analysis of six mitochondrial genes(Maryanovich et al, 2015; Ruggiero et al, 2017). No significant dif-ferences were detected between WT and Efnd1 KO DRGs by thisapproach as well (Fig 3D).

Figure 2. Efhd1 KO neurons exhibit reduced axonalmorphogenesis, neuronal cell loss, and decreasedtarget innervation.(A, B) Dorsal root ganglion (DRG) explants placed in 3Dcollagen from WT and Efhd1 KO E13.5 mouse embryosstained with α-βIII-tubulin. Scale bar: 1,000 μm. (C, D)Limbs from WT and Efhd1 KO E14.5 mouse embryos,stained with α-βIII-tubulin. Arrows point missingaxonal branches in the Efhd1 KO. Scale bar: 500 μm. (E)Axonal lengths of eight WT and eight Efhd1 KO DRGs werequantified by four measurements in threeindependent experiments (N = 3). Graph shows means± SEM (unpaired t test *P = 0.0237). (F) Overall axonalcoverage (total axonal length that cover the limb’ssurface). Graph shows means ± SEM (unpaired t test *P= 0.038). (G) Total number of branching was quantified.Graphs showmeans ± SEM (unpaired t test *P = 0.032).(F, G) Seven WT (N = 7) and eight Efhd1 KO (N = 8)embryos were analyzed in (F, G), data for each embryorepresent the average measurements of both limbs.(H, I) E17.5 WT and Efdh1 KO mouse embryos werestained with anti-Islet1. Scale bar: 100 μm. (J) Graphshows means ± SEM (unpaired t test *P = 0.011). (L, M)E17.5 WT and Efdh1 KO mouse embryos were stainedwith anti-cleaved caspase-3 (CC3). Scale bar: 100 μm. (K)Graph shows means ± SEM (unpaired t test *P =0.020). Arrows point to the CC3-positive cells in theDRGs. (J, K) Five WT (N = 5) and seven Efhd1 KO (N = 7)embryos were analyzed, and the numbers of Islet1-positive neurons (J) and of CC3 positive (K) werequantified in 60 sections/embryo. (N, O) Glabrous skinof adult hind limbs from WT and Efdh1 KO adult micewas stained with α-βIII-tubulin and visualized byconfocal microscopy. Scale bar 100 μm. (P) Nerve-ending fiber quantification. Graph shows means ±SEM (unpaired t test *P = 0.013). 25 sections from eachleft hind limbof threeWT (N = 3) and three Efhd1 KO (N = 3)adult animals were analyzed.


https://doi.org/10.26508/lsa.202000753

In parallel, we examined mitochondria morphology of the abovesamples. The mitochondria in Efhd1 KO growth cones were sig-nificantly shortened compared with those in WT organelles (0.77versus 1.13 μm, respectively) (Fig 3E–I). Together, these resultsdemonstrate that lack of Efhd1 results in metabolic dysfunctions insensory axons, which correlates with reduced axonal morpho-genesis and aberrant mitochondria morphology.

Ablation of Efhd1 causes mitochondrial dysfunction in sensoryneurons

Our observations that Efhd1 KO neurons have similar number ofaxonal mitochondria with irregular morphology prompted us todirectly examine the mitochondrial oxidative phosphorylation(OXPHOS)-mediated ATP production activity in these cells. We usedthe oxygen consumption rate, as measured by the Seahorse XF96system, as a readout of OXPHOS. An equal number of E13.5-dissociated DRGs were plated on PDL/laminin for 4 d, and themitochondrial activity measurements were preformed according to

Styr et al (2019). The Efhd1 KO neurons exhibited clear reduction inthe basal mitochondria respiration (Fig 4A–C), ATP-linked respi-ration, and ATP production (Fig 4A and D). We also detected sig-nificant reduction in the spare respiratory capacity (SRC) based onmulti-comparison Sidak’s test (Fig 4B). However, when the SRC wascalculated according to the baseline of each genotype, we did notdetected any significant difference (Fig 4E). Therefore, we cannotconclude that the SRC is defective in the Efhd1 KO. No differenceswere detected in the non-mitochondrial respiration (proton leak)activity (Fig 4A, B, and F). Overall, our data show that Efhd1 is re-quired for mitochondrial activity and ATP production under basalconditions in sensory neurons.

Last, we tested if the deficits in mitochondrial activity are alsoassociated with changes in mitochondrial motility. Efhd1 KO andWT neurons were plated on PDL/laminin-coated microfluidicschambers for 72 h (Maor-nof et al, 2016). The mitochondria werelabeled by MitoTracker, which localizes to mitochondria regard-less of their membrane potential, and the cells were imaged for 6h (Ionescu et al, 2016). We did not detected any change in the

Figure 3. Efhd1 KO axons exhibit decreased ATP levelsand shortened mitochondria.(A) Soma and axonal ATP levels of cultured dorsal rootganglion neurons were measured using the filter cellculture system. Values are presented as the % of WT.Graph shows means ± SEM of 12 independentexperiments (N = 12) (Wilcoxon signed rank test: soma =NS, axons *P = 0.0034). (B) Schematic illustration ofthe plated dorsal root ganglion explant, axon andgrowth cone (marked by square) from whichmitochondria morphology and number wereanalyzed by TOM20 staining. (C) Quantification of thenumber of mitochondria per growth cones counted inthe super-resolution images. Graphs show means ±SEM of 25 growth cones WT and 20 Efhd1 KO (unpaired ttest NS). (D) Real-time PCR analysis of six mitochondrialgenes (ND1, NADH dehydrogenase subunit 1; ND2,NADH dehydrogenase subunit 2; ND4, NADHdehydrogenase subunit 4; ATP8, ATP synthase protein 8;CYTB1, cytochrome b COX1: cytochrome c oxidasesubunit 1). Graph shows means ± SEM of fourindependent experiments (N = 4) (unpaired t test all NS).(E, F) Wide-field fluorescence image of anti-TOM20staining. Scale bar 2 μm. (E, F, G, H) Super-resolution(STORM) images of anti-TOM20 staining of the samearea of (E, F). Scale bar 2 μm. (I) Quantification ofmitochondria length: each point represents theaverage mitochondria length in a single growth cone;>150 mitochondria were analyzed from eachgenotype, N = 25 WT growth cones and N = 20 Efhd1 KOgrowth cones. Graph shows means ± SEM(Mann–Whitney test: ***P = 0.0002).


https://doi.org/10.26508/lsa.202000753

overall pattern of mitochondrial motility in the Efhd1 KO neurons(Fig 4G–I).

Efhd1 KO sensory neurons have activated metabolic stresssignaling

The mitochondrial dysfunction observed in Efdh1 KO axonsprompted us to examine the activation status of AMPK in thoseneurons. We discovered that, unlike in Lkb1 KOs, AMPK is hyper-phosphorylated in Efhd1 KO axons and DRGs as compared withthe WT control, both in vitro (Figs 5A and C and S5A) and in vivo(Figs 5B and F and S5B). Consistently, we also detected higherphosphorylation of the canonical AMPK target, ACC, in the somaand axons of Efhd1 KO DRGs compared with WT (Figs 5A and D andS5C). As AMPK also regulates the recycling of aberrant mito-chondria via mitophagy through stimulation of Ulk (Toyama et al,2016; Zhang & Lin, 2016), we tested whether this pathway is

enhanced in Efhd1 KOs. In line with AMPK activation, Ulk serine-555 phosphorylation was markedly elevated in the Efhd1 KO DRGsboth in vitro (Figs 5A and E and S5D) and in vivo (Figs 5B and G andS5E).

Next, we tested if there is an increase in autophagic flux in Efhd1KO neurons. We cultured WT and Efhd1 KO DRG neurons in thepresence of the lysosomal inhibitor bafilomycin. Under theseconditions, autophagic vesicles are not degraded, allowing accu-mulation of the lipidated form of the LC3 protein (Zhang et al, 2012;Jiang & Mizushima, 2015; Redmann et al, 2017). Although, as ex-pected, bafilomycin caused an increase in the levels of lipidatedLC3 (LC3-II) in WT neurons, we detected a further increase in LC3-IIlevels in the Efhd1 KO neurons (Fig 5H and I). These results supportthe notion that autophagic flux is enhanced in Efhd1 KO cells,presumably to recruit and recycle the aberrant, dysfunctionalmitochondria. This agrees with our finding that AMPK and Ulk areactivated in Efhd1 KO cells. Overall, these results suggest that Efhd1

Figure 4. Efhd1 KO neurons show decrease inmitochondrial activity.(A, B) Analysis of the oxygen consumption rates (OCRs)in dissociated dorsal root ganglion (DRG) from E13.5 WTand Efhd1 KO embryos by the Seahorse BioscienceXF96 analyzer. (A, B) First, the basal respiration wasmeasured, then (A) oligomycin (1 μM) was added tomeasure the ATP production, and (B) FCCP (4 μM) wasadded to measure the spare respiratory capacity. Therespiration was stopped and non-mitochondrial oxygenconsumption was measured after injection of 0.5 μMrotenone (Rot) and 0.5 μM antimycin A (AA). Data arepresented as the % of the WT baseline (bsln). Theanalysis is based on four independent biologicalexperiments, (N = 4) in each experiment, at least ninewells were used for each condition for each genotype.(A, B) Two-way ANOVA test was performed withSidak’s multi-comparison test: (A) Baseline point 1-2-3,****P ≤ 0.0001, Oligomycin point 1 **P = 0.004, point 2 *P =0.0195 point 3 *P = 0.0462, rotenone and antimycin P =NS; (B) baseline point 1 **P = 0.0041, point 2 *P = 0.0132,point 3 *P = 0.0147, FCCP point 1-2-3 ****P ≤ 0.0001,rotenone and antimycin P = NS. (C, D, E, F) Graphicalquantification of (C) basal respiration, (D) ATPproduction, (E) spare respiratory capacity, and (F) non-mitochondrial ATP production (proton leak) (C:unpaired t test ***P = 0.001, D: unpaired t test **P =0.0042, E, F: unpaired t test NS). Graphs show means ±SEM of OCR that are based on four independentbiological experiments (N = 4), in each experiment, atleast nine wells were used for each condition for eachgenotype. (G, H)Mitochondrial kymograph of axonalmitochondria motility in dissociated neurons fromE13.5 DRGs WT and Efhd1 KO. Scale bar 2 μm. (I)Quantification of the percentage of stationary,anterograde, and retrograde mitochondria in WT andEfhd1 KO neurons. All graphs show mean ± SEM basedon three experiments (N = 3); DRGs were dissectedfrom three WT (N = 3) and three Efhd1 KO (N = 3)embryos (stationary mitochondria: unpaired test NS,anterograde mitochondria: unpaired t test NS,retrograde mitochondria: unpaired t test with Welch’scorrection NS).


https://doi.org/10.26508/lsa.202000753

ablation triggers the activation of metabolic stress pathways in anattempt to compensate for metabolic deficits.

Discussion

In this study, we have identified a metabolic regulator of axonalmorphogenesis. We generated and characterized a mouse modelthat lacks the key metabolic regulator Lkb1. Our data demonstratethat when Lkb1 is ablated in the PNS during early development,neurons display reduced axonal growth both in vivo and in vitro.Cultured Lkb1 KO neurons exhibit a decrease in axonal ATP levels(Fig 6A). Despite this decrease and the consequently elevatedcellular stress level, we did not detect increased phosphorylation ofAMPK, which is considered to be the sensor of cellular stress(Hawley et al, 2003; Hardie, 2007). These data are in line with the

abundance of reports demonstrating that Lkb1 is the principalAMPK kinase (Hawley et al, 2003; Shackelford & Shaw, 2009;Alexander & Walker, 2011).

To identify new regulators of energy homeostasis, we searchedfor alterations of gene expression in Lkb1 KO cells. The expressionof Efhd1 was significantly reduced in the KO neurons. We also foundit to be down-regulated in sensory neurons upon pharmacologicalinhibition of AMPK. These results imply that Efhd1 might be aneffector of the Lkb1–AMPK pathway. Additional studies are requiredto establish if the Lkb1–AMPK pathway directly controls Efhd1 levels,and if so, the exact mode of regulation.

Our results suggest that Efhd1 is required in sensory neurons formitochondrial morphology and function. Interestingly, effects of itsablation are mostly manifested in the axons. This may be due to thefact that axons are more sensitive to mitochondrial dysfunctionbecause of their length or because other metabolic pathways

Figure 5. Efhd1 KO sensory neurons display AMPK andUlk activation along with increased autophagic flux.(A) Immunoblot analysis of AMPK and Ulk activationin E13.5. Dorsal root ganglion (DRG) explants weregrown on the filter cell culture system for 48 h.AMPK (Thr172), ACC (Ser79), and Ulk (Ser555)phosphorylation status were analyzed. (B) AMPK(Thr172) and Ulk (Ser55) phosphorylation statuswere analyzed by immunoblot of in vivo (directlyextracted) E17.5 DRGs. (C, D, E, F, G) The ratio ofphosphorylated to total levels of proteinnormalized to tubulin was quantified using ImageJlevels in E13.5 (C, D, E) and E17.5 (F, G) DRGs wasquantified using ImageJ. (C, D, E) AMPK: soma *P =0.0168, axons **P = 0.0022, (D) ACC: soma **P =0.0069, axons *P = 0.0131, (E) Ulk: t test soma *P =0.0287, axons *P = 0.0185. Graphs show means ±SEM based on three independent experiments (N =3), unpaired t test was used for all. (F, G) AMPK: t test**P = 0.0038 (G) t test ***P = 0.0005. All graphsshow means ± SEM based on N = 4 (p-AMPK) and N =3 (p-Ulk) independent experiments, unpaired t testwas used for both. (H) Immunoblot of LC3I andLC3II levels of WT and Efhd1 KO DRGs treated with

bafilomycin (100 nM) for 4 h. (I) The ratio of LC3-II to tubulin was quantified with ImageJ. Graphs show means ± SEM based on three independent experiments (N = 3)(unpaired t test **P = 0.006).Source data are available for this figure.

Figure 6. Regulation of axonal morphogenesis byEfhd1.(A) Lkb1 controls mitochondrial homeostasis andaxonal ATP production, through Efhd1 and additionalpathways that are required for axonalmorphogenesis. (B) Complete ablation of Efhd1 causesreduction in the axonal ATP levels and mitochondrialabnormalities. This results in the activation of AMPKand Ulk and increased autophagic flux, which iscorrelated with reduced axonal morphogenesis.


https://doi.org/10.26508/lsa.202000753

compensate more efficiently in the soma. Our in vivo analysis alsosupports the idea that axons are more sensitive to the lack of Efhd1,as we detected axonal phenotypic changes as early as E14.5,whereas enhanced cell death was only observed at E17.5.

Our biochemical analysis shows that the Efhd1 KO neurons re-spond to mitochondrial dysfunction by activating metabolic stresspathways, as manifested by the hyper-phosphorylation of AMPK.Although activated AMPK normally rescues the metabolic cellularfunction in stress condition (Gwinn et al, 2008; Toyama et al, 2016),in late stages of DRGs development (E17.5), we observed an in-creased incidence of apoptotic death of Efhd1 KO cells, suggestingthat the energy imbalance at this stage cannot be completelyresolved despite enhanced AMPK activity. Consistent with thehigher apoptotic rate observed at E17.5, innervation of the adultsEfhd1 KO skin was significantly sparser compared with WT.

The presence of aberrantly dysfunctional mitochondria in theEfhd1 KO axons raises the notion that mitophagy might be activatedin Efhd1 KO DRGs. Indeed, we found enhanced Ulk phosphorylationat serine-555, which facilitates mitophagy in response to AMPKactivation and increases the overall autophagic flux (Toyama et al,2016) (Fig 6B). In line with our results, a recent study revealed thatmitochondria shortening and the fission process can be directed byAMPK through the mitochondrial fission factor protein. Once AMPKis activated in response to cellular stress, the mitochondrial fissionfactor is phosphorylated and promotes the constriction and fissionof mitochondria, prompting their engulfment by mitophagosomes(Toyama et al, 2016; Zhang & Lin, 2016).

The exact mitochondrial function of Efhd1 remains to be dis-covered. Previous studies linked Efhd1 to metabolic regulation in thedevelopment of bone marrow cells (Stein et al, 2017) and to the acti-vation ofmitoflashes, short stochastic superoxide bursts during cellularrespiration (Hou et al, 2016). Notably, Efhd1 modulation was describedearlier in a neuronal cell line where Efhd1 overexpression resulted inneurite extension, whereas its down-regulation caused neurite re-duction and subsequent cell death (Tominaga et al, 2006). Given itscalcium-binding capacity (Tominaga et al, 2006), it is tempting tospeculate that Efhd1may serve as anew link between themitochondrialCa2+ and the mitochondria OXPHOS. Multiple studies have clearlydemonstrated that Ca2+ is a physiological regulator of OXPHOS and thatit can facilitate the activation of several enzymes in the tricarboxylic acid(TCA) cycle (Rizzuto et al, 2012). UponCa2+ bindingby its EF-handdomain,Efhd1 may directly interact with proteins of the OXPHOS machinery inthe mitochondrial inner membrane and stimulate their activity. Then,when Efhd1 function is impaired, either partially or completely (as inLkb1 KO and Efhd1 KO, respectively), this Ca2+-dependent regulation ofATP production would be compromised. This may trigger mitochondrialstress due to unmet energy demand during axonal growth, provokingmitochondria recycling through mitophagy.

Despite their sensory neurons’ mitochondrial dysfunction, re-sultant aberrant axonal growth, and neuronal loss, mice with Efhd1deficiency are viable, are fertile, and have normal appearance andbehavior in the cage. Moreover, Efhd1 KO DRGs and axons displayonly limited defects in vitro and in vivo, highlighting the specificcontribution of Efhd1 to axon-remodeling processes but not tobasal homeostasis of the soma.

This study presents a new pathway of mitochondrial regulation.Lkb1, already known as a master metabolic regulator (Shaw et al,

2004; Shackelford & Shaw, 2009; Alexander & Walker, 2011; Pooyaet al, 2014) modulates expression of the mitochondrial proteinEfhd1. As demonstrated herein, Efhd1 has a unique and crucial rolein axonal development of the PNS.

Materials and Methods

Mouse strains: Lkb1 and Efhd1

The Lkb1 conditional KO mouse line was established by crossingLkb1 flox mice (Swisa et al, 2015) with Wnt1–Cre transgenic mice. TheWnt1–Cre transgene is expressed in neural crest cells (progenitorsof [DRG] neurons and glia), and in the midbrain and dorsal neuraltube of the CNS (Charron et al, 2003). Wnt1 expression starts at theembryonic day 8 (E8) in the midbrain and is expressed in its fullpattern by E9 (Charron et al, 2003). The Lkb1 KO mice die shortlyafter birth.

The Efhd1 KO mouse line was generated by CRISPR–cas9 tech-nique. The following two guide RNAs were used to delete a part ofthe Efhd1 gene coding region and its promoter:

Guide 1: sgRNA1-top: CACCgGAGGTTCGCGATCCGGTACG, sgRNA1-bottom: AAACCGTACCGGATCGCGAACCTCc.Guide 2: sgRNA2-top: CACCgGTTCAGTTCGGAGTCCGCGC, sgRNA2-bottom: AAACGCGCGGACTCCGAACTGAACc.

The Efhd1 KO mice are viable and appear indistinguishable fromtheir WT sibs in cage environment. Mice are of mix genetic back-ground in all of the experiments, and we used littermate (WT)controls. Mice were hosed in specific pathogen-free facility, under50% humidity, 12-h light/dark cycle, 22c with standard diet (18%protein, 5% fat). All animal experiments followed protocols ap-proved by The Weizmann Institute of Science Institutional AnimalCare and Use Committee.

Explant culture, medium, and culture method

DRG explants of E13.5 mice were aseptically removed from E13.5embryos in L15 medium (L15 powder [MFCD00217482; Sigma-Aldrich]) dissolved in filtered distilled water (FDW) with 5% fetalbovine serum (10091148; Gibco). Chambers and filter plates: insert(cell culture insert, six-well hanging insert, 1 μm PET, MCRP06H48;Millicell), eight chambers (cell culture slide eight well, 30108; LifeScience Co. Ltd.), six-well plate (140675; Thermo Fisher Scientific),and cell culture dish (430156; Merck). The coating was done for 1 hwith PDL (P6407; Sigma-Aldrich) diluted in FDW, final concentration0.01 mg/ml and, after washing briefly with FDW, laminin (114956-81-9; Sigma-Aldrich), and diluted in filtered F12 (01-095-1A; BiologicalIndustries) for 2 h at 37°C at a final concentration of 10 μg/ml. Theplated explants were grown for 48 h in complete NB (10888022;Thermo Fisher Scientific) supplemented with 2% B27 (17054-044;Thermo Fisher Scientific), 1% penicillin–streptomycin solution(03–0311B; Biological Industries), 1% glutamine (25030081; Gibco),and 12.5 ng/ml NGF (CAS 866405-64-3).

For protein analysis, DRGs were plated in the insert, growth for 48h in NB plus NGF, and the cell bodies and the axons were lysed


https://doi.org/10.26508/lsa.202000753

followed by Western blot of the samples. When indicated, 20 μMcompound C (CAS 866405-64-3; Sigma-Aldrich) for 8 h or 100 nMbafilomycin (LC Laboratories) for 4 h was added after 48 h of cellgrowth in NB. The cultures were then lysed and processed forWestern blot.

For immunofluorescent staining, DRGs were plated in eightchambers or cell culture plates for 48 h in NB plus NGF, fixed with4% formaldehyde for 1 h at room temperature, and stained usingmouse tubulin βIII antibody. Images were taken with DS-Qi2fluorescent microscope, Nikon.

3D collagen cultures

Collagen cultures were performed as previously described (Charronet al, 2003; Romi et al, 2014). Briefly, DRGs, dissected as describedabove, were embedded in 2 mg/ml collagen matrix (1179179001;Roche) supplemented with NB plus NGF. After 5 d in culture, theexplants were fixed with 4% formaldehyde and stained usingmouse tubulin βIII antibody. DRGs were visualized with Leica MZ16Fbinoculars (Nikon), axonal length was measured using ImageJsoftware.

Sample lysis and immunoblot quantification

Two different cell lysis buffers were used for Western blot analysis:

1) RIPA buffer (50mM Tris, pH 7.4, 150mMNaCl, 1%NP40, 0.1% SDS,0.5% deoxycholate, and 1 mM EDTA in double distilled water)supplemented with cOmplete protease inhibitor cocktail(5892791001; Roche), PMSF (CAS 329-98-6; Sigma-Aldrich), andphosphatase inhibitor cocktails 1 and 2 (P5726; Sigma-Aldrich).

2) Triton-lysis buffer (150 mM NaCl, 10% glycerol, 1% Triton X-100,and 10mM Tris, pH 7.4, in double distilled water) supplementedwith cOmplete protease inhibitor cocktail, PMSF, and phos-phatase inhibitor cocktails 1 and 2.

All the samples were lysed in RIPA buffer (1), except the samplesfrom E13.5 embryos used to detect P-Ulk ser 555/Ulk tot, P-ACC/ACCtot, for which the lysis was performed with lysis buffer (2). Im-munoblots for P-AMPK, P-Ulk, and LC3 I-II at E13.5 and P-Ulk andP-AMPK at E17.5 were loading-normalized by re-probing the originalmembranes with the respective loading control (tubulin βIII orβ-actin) and total protein antibodies (total AMPK, Ulk, and tubulinβIII for LC3). The immunoblots were scanned and analyzed usingImageJ program. The samples for assessment of P-ACC/ACC at E13.5were split and loaded on two pairs of identical gels. The intensitiesof the phosphorylated and the total forms detected on the first setof gels were then normalized on the tubulin βIII or β-actin asdetected on the corresponding replica gel set, and then the valuesof the phosphorylated form/tubulin βIII or phosphorylated form/β-actin ratios were further adjusted relative to the ratios of total/tubulinβIII or total/β-actin.

Immunostaining

After fixation with 4% formaldehyde. DRGs placed in the collagenmatrix were gently washed with PBS. Blocking was donewith 3%BSA

(160069; MP Biomedicals, LLC) and 0.1% Triton X-100 (CAS900-93-1;Sigma-Aldrich) in PBS for 1 h. The blocking solution was used forincubation with primary and secondary antibodies. PBS washeswere made between each step.

Whole-mount embryo limbs

Limbs of E13.5 and E14.5 embryos were stained with anti-tubulin βIIIantibody and anti-neurofilament antibody 2H3 following the iDiscomethodology (Renier et al, 2014). For total axonal coverage andbranching quantification, the NeuroMath software was used (Rishalet al, 2013). The numbers of samples processed in each experimentare specified in the corresponding figure legends.

Gene expression microarray

DRGs from E13.5 Lkb1 KO and WT embryos were harvested and RNAextracted as described in Rio et al (2010). For each sample, wepooled DRGs from two embryos. Overall, we analyzed the data from10 microarrays, profiling five Lkb1 KO and five WT samples. cDNAlibrary construction and microarray hybridization were performedat the microarray unit of the Weizmann Institute of Science. Datawere analyzed using MATLAB, false discovery rate 0.01.

Data are available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE146756.

Immunohistochemistry of mouse embryos sections

E15.5–E17.5 embryos were fixed for 24 h in 4% formaldehyde andstained as described in Maor-nof et al (2016) with anti-Islet andanti-cleaved caspase-3 antibody. Numbers of embryos and sec-tions used in individual experiments are specified in the figurelegends.

The number of Iset1-positive neurons per DRG was calculated bycomputational approach as described in detail in Maor-nof et al(2016). The number of cleaved caspase-3–positive neurons wasmanually counted in ImageJ.

Adult skin innervation

The hind limb glabrous skin of 2-mo-old mice was stained asdescribed in Marvaldi et al (2015) (Zylka et al, 2005). For the nervequantification, the fiber number per 150 μm in the selected epi-dermis area was counted as described in Marvaldi et al (2015). Thenumbers of mice, limbs, and sections analyzed are specified in thefigure legends.

ATP measurement

Whole explants or axonal samples derived from Lkb1 KO, Efhd1 KO,and WT inserts were collected in 100 mM Tri and 4 mM EDTA, pH 7.75,and incubated for 2 min at 90°C. ATP concentration was measuredusing ATP Bioluminescence Assay Kit CLS II (11699695001; Roche).Whole protein level in the same samples were quantified (BCAprotein assay kit; Pierce). The ATP measurements were then nor-malized to the protein concentration.


https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE146756https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE146756https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE146756https://doi.org/10.26508/lsa.202000753

Antibodies

Anti-tubulin βIII antibody (clone Tuj1; R&D Systems), at 1:20,000 forWestern blot, 1:1,000 for immunoistochemistry (IHC) staining.P-AMPK thr172: phospho-AMPK-α (Thr172) (40H9) Cell Signaling, at 1:1,000. AMPK: AMPK-α (23A3) Cell Signaling, at 1:1,000. P-ACC ser 79:phospho-ACC (Ser79) Cell Signaling, at 1:1,000. ACC: anti-acetylcoenzyme A carboxylase antibody [EP687Y], AB-ab45174 Abcam,at 1:2,000. P-ULK ser555: phospho-ULK1 (Ser555) (D1H4) Cell Sig-naling, at 1:1,000. Caspase-3-cleaved: cleaved caspase-3 (Asp175)Cell Signaling, at 1:200. Islet antibody: Developmental Studies Hy-bridoma bank 39.3F7, at 1:200. Tom20: Santa Cruz Tom20 (FL-145): sc11415, at 1:1,000 for immunofluorescence (STORM analysis) andTOM20 (D8T4N) Cell Signaling at 1:1,000 for immunoblot. β-Actin:β-Actin (13E5) Cell Signaling, at 1:20,000. Efhd1 antibody was a kindgift from the laboratory of Yasuhiro Tomooka (Tokyo University ofScience), at 1:1,000. Anti-neurofilament antibody 2H3, Develop-mental Studies Hybridoma bank, at 1:200.

Antimouse and antirabbit antibodies conjugated with Alexa 549,Alexa 488, or Alexa 647 fluorophores were used at 1:200 (JacksonImmunoResearch Laboratories). Secondary antibodies for Westernblotting: goat antimouse IgG-HRP (JIR 155-035) and goat antirabbitIgG-HRP (JIR 111-035) from Jackson, both at 1:5,000.

Stochastic optical reconstruction microscopy (STORM) imaging

DRGs from E13.5 embryos were plated on cell culture dish as de-scribed above.

Three-dimensional super-resolution images were recordedusing a Vutara SR200 STORM microscope (Bruker) based on single-molecule localization biplane technology with 60× Olympus water-immersion objective (1.2 NA). Mitochondria (anti-TOM20 staining)labeled with AlexaFluor 647 were imaged using 640-nm excitationlaser and 405-nm activation laser in an imaging buffer composed of5 mM cysteamine, oxygen scavengers (7 μM glucose oxidase and 56nM catalase) in 50 mM Tris with 10 mM NaCl, and 10% glucose at pH8.0. Images were recorded using Evolve 512 EMCCD camera (Pho-tometrics) with gain set at 50, frame rate at 50 Hz, and maximalpower of 640 and 405 nm lasers set at 6 and 0.05 kW/cm2, re-spectively. The total number of frames acquired was typically15,000. Data analysis was performed using Vutara SRX software,localized particles were subjected to threshold value (set to 5) thatis defined in Vutara SRX as standard deviations above the framebackground value, which is determined based on themean value ofthe border pixels in each frame. Mitochondrial length wasmanuallyanalyzed using the SRX Vutara Software.

Mitochondrial DNA quantification

DRGs were cultured in six-well chambers as described above. Onthe second day, the cultures were treated with FUDR (5-fluoro-29-deoxyuridine, CAS 50-91-9, 100 nM; Merck) for 8 h. On the third day,FUDR was removed. On the fifth day, DNA was extracted with theMasterPure DNA purification kit (Cat. no. MCD85201; Epicentre), andquantitative real-time PCR was performed using SYBR Green (Cat. no.4385612; Applied Biosystems). Expression levelswere determinedusing

the comparative cycle threshold (2−ΔΔCt) method. 18S ribosomal RNAserved as housekeeping genes.

Primers used were as follows:

ND1: FWD: 59-TGCACCTACCCTATCACTCA-39REV: 59-GCTCATCCTGATCATAGAATGG-39ND1 and ND4: FWD: 59-CACTAATGCTACTACCACTAACCTGACTATC-39REV: 59-TGTCATAGAAGTGTTAGGCTGGTTAA AC-39Cytb1: FWD: 59-ACG TCC TTC CAT GAG GAC AA-39REV: 59-GAG GTG AAC GAT TGC TAG GG-39COX1: FWD: 59-GCCTTTGCTTCAAAACGAGA-39REV: 59-GGTTGGTTCCTCGAATGTGT-39MT-ATP8: FWD: 59-GCCACAACTAGATACATCAACATGA-39REV: 59-GGTTGTTAGTGATTTTGGTGAAGGT-3918S: FWD-59-AAACGGCTACCACATCCAAG-39REV-59-CCTCCAATGGATCCTCGTTA-39

Mitochondrial respiration studies (Seahorse)

For the mitochondrial respiration studies, DRGs were dissected asdescribed above, washed in HBSS, and dissociated by incubation intrypsin–EDTA solution B (03-052-1B; Biological Sciences) for 5 min.After trypsinization, dissociated DRGs were suspended in L15 plusserum and plated at a concentration of 30 × 104 on PDL/laminin(described above)–coated XF 96 plates (102601-100; Agilent) withneurobasal (described before). On the second day, FUDR (100 nm)was added to the media for 4 h to eliminate dividing cells, thusobtaining a purer neuronal culture. On day 4, mitochondrial res-piration was examined using the Seahorse XF96 analyzer (Agilent)and the XF Cell Mito Stress Test Kit according to the manufacturer’sinstructions and as described in Karkucinska-Wieckowska et al(2015), Ruggiero et al (2017), Styr et al (2019). The analysis wasperformed in two parallel experiments as described in Styr et al.,2019. Briefly, in the first, respiration was measured under basalconditions and in response to the ATP synthases inhibitor oligo-mycin (1 μM), whereas in the second, respiration was measuredunder basal condition and in the presence of the electron transportchain accelerator ionophore FCCP (trifluorocarbonylcyanidephenylhydrazone, 4 μM). In both experiments, respiration wasstopped by addition of the electron transport chain inhibitorsrotenone and antimycin A (both at 0.5 μM). The neurons were thenfixed with 4% formaldehyde and stained with tubulin βIII antibody.The plates were imaged, and the number of neurons was recal-culated. Respiration values were normalized to the number ofneurons. For each condition (oligomycin and rotenone/antimycin,and FCCP and rotenone/antimycin), we preformed four indepen-dent experiments; at least nine wells were used for each genotype,in each experiment. The baseline graph was generated by analyzingthe third value of the baseline, in all the experiments. The ATPproduction graph was generated by subtracting the first valueafter oligomycin treatment from the third value of the baseline, inall the experiments. The SRC graph was generated by subtractingthe third baseline value from the first value after FCCP injection, inall the experiments. The proton leak graph was generated bysubtracting the first value after the injection of rotenone andantimycin from the third value after oligomycin injection, in all theexperiments.


https://doi.org/10.26508/lsa.202000753

Mitochondrial motility

E13.5-dissociated DRGs (described above) from Lkb1 and Efhd1 KOsand their respective control neurons were plated on PDL/laminin–coated microfluidic chambers. After 72 h, the mitochondriawere labeled by 5 nM TMRE (tetramethylrhodamine ethyl ester;Thermo Fisher Scientific) in the Lkb1 KO experiments and Mito-Tracker (M7514; Thermo Fisher Scientific) in the Efhd1 KO experi-ments and imaged for 6 h. Motility analyses were performed byImageJ as described in Ionescu et al (2016).

Statistical analysis

Statistical analysis was performed using GraphPad Prism 7.0software, Mean ± SEM is presented. The number of experiments (N)is indicated the figure legends. Normality of the data was definedusing the Shapiro–Wilk normality test. For non-normally distributeddata, Wilcoxon signed rank test and Mann–Whitney test wereperformed. For normally distributed homoscedastic data. two-tailed Student’s t test was used and for non-homoscedastic ttest with Welch’s correction. For the Seahorse analysis, two-wayANOVA test and Sidak’s multi-comparison test were performed.

Supplementary Information

Supplementary Information is available at https://doi.org/10.26508/lsa.202000753.

Acknowledgements

We thank the Yaron lab members for advice and criticism, Eran Perlson andTal Pery Gradus for the mitochondrial motility analysis, Ruggiero Antonellafor consultation and advise on the Seahorse experiments and analysis,Vladimir Kiss for the help with the confocal microscope, Ron Rotkopf forexcellent statistical assistance, and Andrew Kovalenko for critically readingthe manuscript. This work was supported by funding to A Yaron from TheLegacy Heritage Biomedical Science Partnership of the Israel ScienceFoundation (1004/09), National Science Foundation (NSF-IOS/BOI grant[1556968]), and The Nella and Leon Benoziyo Center for Neurological Dis-eases; Mr and Mrs James Kelly, Helene T Fox, Daniel C Andreae, the SamuelAba, and Sisel Klurman Foundation in honor of Prof. Joel Sussman; theAdvantage Trust in honor of Prof. Joel Sussman; and the Estate of Ethel LenaLevy. A Yaron is an incumbent of the Jack and Simon Djanogly ProfessorialChair in Biochemistry.

Author Contributions

V Ulisse: conceptualization, data curation, investigation, method-ology, and writing—original draft, review, and editing.S Dey: data curation and formal analysis.DE Rothbard: data curation and investigation.E Zeevi: data curation and investigation.I Gokhman: data curation and investigation.T Dadosh: data curation and investigation.A Minis: conceptualization, data curation, and formal analysis.

A Yaron: conceptualization, resources, supervision, funding ac-quisition, project administration, and writing—original draft, review,and editing.

Conflict of Interest Statement

The authors declare that they have no conflict of interest.

References

Alexander A, Walker CL (2011) The role of LKB1 and AMPK in cellular responsesto stress and damage. FEBS Lett 585: 952–957. doi:10.1016/j.febslet.2011.03.010

Barnes AP, Lilley BN, Pan YA, Plummer LJ, Powell AW, Raines AN, Sanes JR,Polleux F (2007) LKB1 and SAD kinases define a pathway required forthe polarization of cortical neurons. Cell 129: 549–563. doi:10.1016/j.cell.2007.03.025

Charron F, Stein E, Jeong J, McMahon AP, Tessier-Lavigne M (2003) Themorphogen sonic hedgehog is an axonal chemoattractant thatcollaborates with Netrin-1 in midline axon guidance. Cell 113: 11–23.doi:10.1016/s0092-8674(03)00199-5

Courchet J, Lewis TL, Lee S, Courchet V, Liou DY, Aizawa S, Polleux F (2013)XTerminal axon branching is regulated by the LKB1-NUAK1 kinasepathway via presynaptic mitochondrial capture. Cell 153: 1510–1525.doi:10.1016/j.cell.2013.05.021

Germain M, Nguyen MG, Khacho M, Patten DA, Screaton RA, Park DS, Slack RS(2013) LKB1-regulated adaptive mechanisms are essential forneuronal survival following mitochondrial dysfunction. Hum MolGenet 22: 952–962. doi:10.1093/hmg/dds500

Gwinn DM, Shackelford DB, Egan DF, Mihaylova MM, Mery A, Vasquez DS, TurkBE, Shaw RJ (2008) AMPK phosphorylation of raptor mediates ametabolic checkpoint. Mol Cell 30: 214–226. doi:10.1016/j.molcel.2008.03.003

Hardie DG (2007) AMP-activated/SNF1 protein kinases: Conserved guardiansof cellular energy. Nat Rev Mol Cell Biol 8: 774–785. doi:10.1038/nrm2249

Hardie DG (2011) AMP-activated protein kinase-an energy sensor thatregulates all aspects of cell function. Genes Dev 25: 1895–1908.doi:10.1101/gad.17420111

Hawley SA, Boudeau J, Reid JL, Mustard KJ, Udd L, Mäkelä TP, Alessi DR, HardieDG (2003) Complexes between the LKB1 tumor suppressor, STRADalpha/beta and MO25 alpha/beta are upstream kinases in the AMP-activated protein kinase cascade. J Biol 2: 28. doi:10.1186/1475-4924-2-28

Hou T, Jian C, Xu J, Huang AY, Xi J, Hu K, Wei L, Cheng H, Wang X (2016)Identification of EFHD1 as a novel Ca2+ sensor for mitoflash activation.Cell Calcium 59: 262–270. doi:10.1016/j.ceca.2016.03.002

Ionescu A, Zahavi EE, Gradus T, Ben-Yaakov K, Perlson E (2016)Compartmental microfluidic system for studying muscle-neuroncommunication and neuromuscular junction maintenance. Eur J CellBiol 95: 69–88. doi:10.1016/j.ejcb.2015.11.004

Jiang P, Mizushima N (2015) LC3- and p62-based biochemical methods for theanalysis of autophagy progression in mammalian cells. Methods 75:13–18. doi:10.1016/j.ymeth.2014.11.021

Karkucinska-Wieckowska A, Pronicki M, Wieckowski MR (2015) Histoenzymaticmethods for visualization of the activity of individual mitochondrialrespiratory chain complexes in the muscle biopsies from patientswith mitochondrial defects.Methods Mol Biol 1241: 85–93. doi:10.1007/978-1-4939-1875-1_8


https://doi.org/10.26508/lsa.202000753https://doi.org/10.26508/lsa.202000753https://doi.org/10.1016/j.febslet.2011.03.010https://doi.org/10.1016/j.febslet.2011.03.010https://doi.org/10.1016/j.cell.2007.03.025https://doi.org/10.1016/j.cell.2007.03.025https://doi.org/10.1016/s0092-8674(03)00199-5https://doi.org/10.1016/j.cell.2013.05.021https://doi.org/10.1093/hmg/dds500https://doi.org/10.1016/j.molcel.2008.03.003https://doi.org/10.1016/j.molcel.2008.03.003https://doi.org/10.1038/nrm2249https://doi.org/10.1038/nrm2249https://doi.org/10.1101/gad.17420111https://doi.org/10.1186/1475-4924-2-28https://doi.org/10.1186/1475-4924-2-28https://doi.org/10.1016/j.ceca.2016.03.002https://doi.org/10.1016/j.ejcb.2015.11.004https://doi.org/10.1016/j.ymeth.2014.11.021https://doi.org/10.1007/978-1-4939-1875-1_8https://doi.org/10.1007/978-1-4939-1875-1_8https://doi.org/10.26508/lsa.202000753

Lilley BN, Pan YA, Sanes JR (2013) SAD kinases sculpt axonal arbors of sensoryneurons through long- and short-term responses to neurotrophinsignals. Neuron 79: 39–53. doi:10.1016/j.neuron.2013.05.017

Maor-nof M, Romi E, Shalom HS, Ulisse V, Raanan C, Nof A, Leshkowitz D, LangR, Yaron A (2016) Axonal degeneration is regulated by atranscriptional program that coordinates expression of pro- and anti-degenerative factors. Neuron 92: 991–1006. doi:10.1016/j.neuron.2016.10.061

Marvaldi L, Thongrong S, Kozłowska A, Irschick R, Pritz CO, Bäumer B, RonchiG, Geuna S, Hausott B, Klimaschewski L (2015) Enhanced axonoutgrowth and improved long-distance axon regeneration insprouty2 deficient mice. Dev Neurobiol 75: 217–231. doi:10.1002/dneu.22224

Maryanovich M, Zaltsman Y, Ruggiero A, Goldman A, Shachnai L, Zaidman SL,Porat Z, Golan K, Lapidot T, Gross A (2015) An MTCH2 pathwayrepressing mitochondria metabolism regulates haematopoietic stemcell fate. Nat Commun 6: 7901. doi:10.1038/ncomms8901

Misgeld T, Schwarz TL (2017) Mitostasis in neurons: Maintaining mitochondriain an extended cellular architecture. Neuron 96: 651–666. doi:10.1016/j.neuron.2017.09.055

Nakano A, Takashima S (2012) LKB1 and AMP-activated protein kinase:Regulators of cell polarity. Genes Cells 17: 737–747. doi:10.1111/j.1365-2443.2012.01629.x

Pooya S, Liu X, Kumar VBS, Anderson J, Imai F, Zhang W, Ciraolo G, Ratner N,Setchell KDR, Yutaka Y, et al (2014) The tumour suppressor LKB1regulates myelination through mitochondrial metabolism. NatCommun 5: 4993. doi:10.1038/ncomms5993

Redmann M, Benavides GA, Berryhill TF, Wani WY, Ouyang X, Johnson MS, RaviS, Barnes S, Darley-Usmar VM, Zhang J (2017) Inhibition of autophagywith bafilomycin and chloroquine decreases mitochondrial qualityand bioenergetic function in primary neurons. Redox Biol 11: 73–81.doi:10.1016/j.redox.2016.11.004

Renier N, Wu Z, Simon DJ, Yang J, Ariel P, Tessier-Lavigne M (2014) IDISCO: Asimple, rapid method to immunolabel large tissue samples forvolume imaging. Cell 159: 896–910. doi:10.1016/j.cell.2014.10.010

Rio DC, Are M, Hannon GJ, Nilsen TW (2010) Purification of RNA using TRIzol (TRIreagent). Cold Spring Harbor Protoc 5: 1–4. doi:10.1101/pdb.prot5439

Rishal I, Golani O, Rajman M, Costa B, Ben-Yaakov K, Schoenmann Z, Yaron A,Basri R, Fainzilber M, Galun M (2013) WIS-neuromath enables versatilehigh throughput analyses of neuronal processes. Dev Neurobiol 73:247–256. doi:10.1002/dneu.22061

Rizzuto R, De Stefani D, Raffaello A, Mammucari C (2012) Mitochondria assensors and regulators of calcium signalling. Nat Rev Mol Cell Biol 13:566–578. doi:10.1038/nrm3412

Romi E, Gokhman I, Wong E, Antonovsky N, Ludwig A, Sagi I, Saftig P, Tessier-Lavigne M, Yaron A (2014) ADAM metalloproteases promote adevelopmental switch in responsiveness to the axonal repellantSema3A. Nat Commun 5: 1–15. doi:10.1038/ncomms5058

Ruggiero A, Aloni E, Korkotian E, Zaltsman Y, Oni-Biton E, Kuperman Y, TsooryM, Shachnai L, Levin-Zaidman S, Brenner O, et al (2017) Loss offorebrain MTCH2 decreases mitochondria motility and calciumhandling and impairs hippocampal-dependent cognitive functions.Sci Rep 7: 1–13. doi:10.1038/srep44401

Shackelford DB, Shaw RJ (2009) The LKB1-AMPK pathway: Metabolism andgrowth control in tumour suppression. Nat Rev Cancer 9: 563–575.doi:10.1038/nrc2676

Shaw RJ, Kosmatka M, Bardeesy N, Hurley RL, Witters LA, DePinho RA, CantleyLC (2004) The tumor suppressor LKB1 kinase directly activates

AMP-activated kinase and regulates apoptosis in response to energystress. Proc Natl Acad Sci U S A 101: 3329–3335. doi:10.1073/pnas.0308061100

Shelly M, Cancedda L, Heilshorn S, Sumbre G, Poo MM (2007) LKB1/STRADpromotes axon initiation during neuronal polarization. Cell 129:565–577. doi:10.1016/j.cell.2007.04.012

Sheng ZH (2017) The interplay of axonal energy homeostasis andmitochondrial trafficking and anchoring. Trends Cell Biol 27: 403–416.doi:10.1016/j.tcb.2017.01.005

Spillane M, Ketschek A, Merianda TT, Twiss JL, Gallo G (2013) Mitochondriacoordinate sites of axon branching through localized intra-axonalprotein synthesis. Cell Rep 5: 1564–1575. doi:10.1016/j.celrep.2013.11.022

Stein M, Dütting S, Mougiakakos D, Bösl M, Fritsch K, Reimer D, Urbanczyk S,Steinmetz T, SchuhW, Bozec A, et al (2017) A definedmetabolic state inpre B cells governs B-cell development and is counterbalanced bySwiprosin-2/EFhd1. Cell Death Differ 24: 1239–1252. doi:10.1038/cdd.2017.52

Styr B, Gonen N, Zarhin D, Ruggiero A, Atsmon R, Gazit N, Braun G, Frere S,Vertkin I, Shapira I, et al (2019) Mitochondrial regulation of thehippocampal firing rate set point and seizure susceptibility. Neuron102: 1009–1024.e8. doi:10.1016/j.neuron.2019.03.04

Swisa A, Granot Z, Tamarina N, Sayers S, Bardeesy N, Philipson L, Hodson DJ,Wikstrom JD, Rutter GA, Leibowitz G, et al (2015) Loss of liver kinase B1(LKB1) in beta cells enhances glucose-stimulated insulin secretiondespite profound mitochondrial defects. J Biol Chem 290:20934–20946. doi:10.1074/jbc.m115.639237

Tominaga M, Kurihara H, Honda S, Amakawa G, Sakai T, Tomooka Y (2006)Molecular characterization of mitocalcin, a novel mitochondrial Ca2+-binding protein with EF-hand and coiled-coil domains. J Neurochem96: 292–304. doi:10.1111/j.1471-4159.2005.03554.x

Toyama EQ, Herzig S, Courchet J, Lewis TL, Jr, Losón OC, Hellberg K, Young NP,Chen H, Polleux F, Chan DC, et al (2016) AMP-activated protein kinasemediates mitochondrial fission in response to energy stress. Science351: 275–281. doi:10.1126/science.aab4138

Tymanskyj SR, Yang BH, Verhey KJ, Ma L (2018) MAP7 regulates axonmorphogenesis by recruiting kinesin-1 to microtubules andmodulating organelle transport. Elife 7: 1–27. doi:10.7554/elife.36374

Vaarmann A, Mandel M, Zeb A, Wareski P, Liiv J, Kuum M, Antsov E, Liiv M,Cagalinec M, Choubey V, et al (2016) Mitochondrial biogenesis isrequired for axonal growth. Development 143: 1981–1992. doi:10.1242/dev.128926

Zhang X, Garbett K, Veeraraghavalu K, Wilburn B, Gilmore R, Mirnics K, SisodiaSS (2012) A role for presenilins in autophagy revisited: Normalacidification of lysosomes in cells lacking PSEN1 and PSEN2. J Neurosci32: 8633–8648. doi:10.1523/jneurosci.0556-12.2012

Zhang CS, Lin SC (2016) AMPK promotes autophagy by facilitatingmitochondrial fission. Cell Metab 23: 399–401. doi:10.1016/j.cmet.2016.02.017

Zylka MJ, Rice FL, Anderson DJ (2005) Topographically distinct epidermalnociceptive circuits revealed by axonal tracers targeted to Mrgprd.Neuron 45: 17–25. doi:10.1016/j.neuron.2004.12.015

License: This article is available under a CreativeCommons License (Attribution 4.0 International, asdescribed at https://creativecommons.org/licenses/by/4.0/).


https://doi.org/10.1016/j.neuron.2013.05.017https://doi.org/10.1016/j.neuron.2016.10.061https://doi.org/10.1016/j.neuron.2016.10.061https://doi.org/10.1002/dneu.22224https://doi.org/10.1002/dneu.22224https://doi.org/10.1038/ncomms8901https://doi.org/10.1016/j.neuron.2017.09.055https://doi.org/10.1016/j.neuron.2017.09.055https://doi.org/10.1111/j.1365-2443.2012.01629.xhttps://doi.org/10.1111/j.1365-2443.2012.01629.xhttps://doi.org/10.1038/ncomms5993https://doi.org/10.1016/j.redox.2016.11.004https://doi.org/10.1016/j.cell.2014.10.010https://doi.org/10.1101/pdb.prot5439https://doi.org/10.1002/dneu.22061https://doi.org/10.1038/nrm3412https://doi.org/10.1038/ncomms5058https://doi.org/10.1038/srep44401https://doi.org/10.1038/nrc2676https://doi.org/10.1073/pnas.0308061100https://doi.org/10.1073/pnas.0308061100https://doi.org/10.1016/j.cell.2007.04.012https://doi.org/10.1016/j.tcb.2017.01.005https://doi.org/10.1016/j.celrep.2013.11.022https://doi.org/10.1038/cdd.2017.52https://doi.org/10.1038/cdd.2017.52https://doi.org/10.1016/j.neuron.2019.03.04https://doi.org/10.1074/jbc.m115.639237https://doi.org/10.1111/j.1471-4159.2005.03554.xhttps://doi.org/10.1126/science.aab4138https://doi.org/10.7554/elife.36374https://doi.org/10.1242/dev.128926https://doi.org/10.1242/dev.128926https://doi.org/10.1523/jneurosci.0556-12.2012https://doi.org/10.1016/j.cmet.2016.02.017https://doi.org/10.1016/j.cmet.2016.02.017https://doi.org/10.1016/j.neuron.2004.12.015https://creativecommons.org/licenses/by/4.0/https://creativecommons.org/licenses/by/4.0/https://doi.org/10.26508/lsa.202000753

Regulation of axonal morphogenesis by the mitochondrial protein Efhd1IntroductionResultsLkb1 KO sensory neurons display normal polarization but reduced axonal growth in vitroLkb1 KO sensory neurons exhibit reduced axonal morphogenesis in vivoLkb1 KO neurons display reduced axonal ATP levelsLkb1-deficient sensory axons display normal mitochondria motilityLkb1 KO sensory neurons exhibit reduced expression of the mitochondrial protein Efhd1Efhd1 is required for axonal growth in vitroEfhd1 KO sensory neurons display aberrant axonal development and increased cell death in vivoEfhd1 KO axons manifest lower ATP levels and shortened mitochondriaAblation of Efhd1 causes mitochondrial dysfunction in sensory neuronsEfhd1 KO sensory neurons have activated metabolic stress signaling

DiscussionMaterials and MethodsMouse strains: Lkb1 and Efhd1Explant culture, medium, and culture method3D collagen culturesSample lysis and immunoblot quantificationImmunostainingWhole-mount embryo limbsGene expression microarrayImmunohistochemistry of mouse embryos sectionsAdult skin innervationATP measurementAntibodiesStochastic optical reconstruction microscopy (STORM) imagingMitochondrial DNA quantificationMitochondrial respiration studies (Seahorse)Mitochondrial motilityStatistical analysis

Supplementary InformationAcknowledgementsAuthor ContributionsConflict of Interest StatementAlexander A, Walker CL (2011) The role of LKB1 and AMPK in cellular responses to stress and damage. FEBS Lett 585: 952–957. ...

regulation of axonal morphogenesis by the mitochondrial ...€¦ · function of efhd1 in sensory...

Documents